Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

DaFranker comments on Politics is the Mind-Killer - Less Wrong

72 Post author: Eliezer_Yudkowsky 18 February 2007 09:23PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (228)

Sort By: Old

You are viewing a single comment's thread. Show more comments above.

Comment author: DaFranker 30 October 2012 05:08:18PM *  -1 points [-]

The difficult part about finding the optimal perfect-rationalist "right answer" for things related to politics is that politics is like an exceptionally difficult, complex and heavy computer program currently being coded by hundreds of programmers, most of which have no formal Computer Science education, and then managing to produce optimal software out of it with only the help of two or three of those coders - the best possible program that achieves absolutely everything that the client wants in exactly the best possible way.

Unfortunately, the example program is so complex that near-optimal solutions do not converge towards the same location in the conceptspace of possible programs, and each programmer has his own idea of what might be good, so you have a large multitude of possible local maximums, all of which are of unknown order of magnitude (let alone being able to decide which is better) and unknown cost (and you can rule out perfect cost-effectiveness calculations), and often even with unclear value-of-information that varies across conceptspace function of the properties of this area of conceptspace (e.g. it has a higher expected human-values cost to experiment with totalitarian-like forms of government than with democratic ones, for a vague picture).

Overall, not only is there a ton of biases, but information is costly and the space of possibilities is vast, and the near-optimals or optimization candidates / hypotheses are not condensed or sometimes not even remotely near eachother. Thus, discussing politics rationally isn't just difficult here - politics are a set (space? field?) of complex Hard problems with tons of data, variables and unknowns, and would probably still be among the more difficult problems to solve if all humans were suddenly replaced with perfect bayesian agents.

Comment author: gokhalea 30 October 2012 09:05:38PM 1 point [-]

Thanks, I agree with nearly all you points but want to push on a particular point you made: (btw, how do you guys have that blue line to show you are responding to a particular comment??):

"Thus, discussing politics rationally isn't just difficult here - politics are a set (space? field?) of complex Hard problems with tons of data, variables and unknowns, and would probably still be among the more difficult problems to solve if all humans were suddenly replaced with perfect bayesian agents."

I would argue that politics is difficult to rationalize BECAUSE politics are in a separate space/field. In other words, i think discussing politics rationally in a manner consistent with Less Wrong's definition of rationality (see "what we mean by rationality" article) is impractical and does not further any knowledge because the definition simply does not apply in a way it can apply to other areas discussed here. Going "funny in the head" is not the reason we cannot apply rationality to politics, we go "funny in the head" because we are using a model that does not work -- we are trying to find answers to questions that, as you describe, are subject to so much uncertainty we are forced to resort to biases. We fail to consider the possibility that there is no right answer -- for those that argue that there is an answer, but humans can't reach it (a HUGELY convenient position) -- that is the same thing, practically speaking, as not having an answer:

If the problem is the model, not the people, change the mode to one where the search is not for the right answer, but a deep understanding of why particular people have viewpoints and the relative arguments therefor. Sure, its not an "answer" to how the world is (or should be), but its a huge step forward in understanding how the world works -- a noble goal if you ask me. The current model of rationality used here simply doesn't allow for this. We are obsessed with certainty, even when there is more value to be derived from better understanding the relative uncertainty.

In his article on rationalization (contrasting it with rationality), Eliezer says: ""Rationalization" is a backward flow from conclusion to selected evidence. First you write down the bottom line, which is known and fixed; the purpose of your processing is to find out which arguments you should write down on the lines above. This, not the bottom line, is the variable unknown to the running process."

On a most general level, it seems the very definition of "rationality", requiring a normative conclusion, is a result of rationalization. More specifically, saying "politics is a mind killer" to avoid applying rationality to politics, and then telling us why people are flawed and can't analyze these things also sounds a lot like rationalization. Is that a forward flowing, rational conclusion? No one here can or will apply rationality in coming to political conclusions (whether a firm answer or not) -- so how can you tell me that its a mind-killer? Perhaps politics is not a mind-killer and instead, politics, within a restrictive definition of rationality, is a mind-killer. These are not fighting words. I just want to understand.

Comment author: Vaniver 30 October 2012 09:21:00PM *  1 point [-]

(btw, how do you guys have that blue line to show you are responding to a particular comment??

If you begin a paragraph with >, it will put it in block quote format.

Comment author: NancyLebovitz 30 October 2012 10:35:14PM 1 point [-]

It really is blue-- I'd been assuming it was black. Did it used to be black?

Comment author: Vaniver 30 October 2012 11:46:26PM 0 points [-]

I recall it as being blue since I arrived (getting close to two years ago). I have not paid close attention before now.

Comment author: shminux 31 October 2012 12:05:23AM 4 points [-]

Maybe you missed EY's point, or maybe I'm missing yours. Politics definitely can be discussed rationally, but it is really really hard to keep your identity small while doing so. Every participant in a political discussion has to be constantly aware of their own emotions fueled by a cached arguments associated with a specific wing/party/position, and be skilled at modeling how potential readers would inadvertently misinterpret one's statement, causing them emotional upheaval. And it only takes a small misstep to get people riled up about the issue.

The rule of thumb is "if you identify with any political party/group/movement, you are not qualified to have a rational discourse about politics". Example: if you want to start your reply with "as a libertarian, I ...", you have failed. Another example of a false start: "Republicans do not understand that ..."

Comment author: chaosmosis 31 October 2012 12:46:33AM 1 point [-]

Gokhalea's point is largely that political analyses are so difficult that attempting to apply rationality to them will still produce biased and nonsensical results. You didn't really address that.

Gokhalea, I agree that rational analyses of politics are difficult. You seem to believe that they're functionally impossible. Can you explain why? Also, I don't understand why you feel that avoiding politics on LessWrong is a form of rationalization. What's motivating this rationalization? Finally, I don't understand why you feel that a model of politics which seeks to understand different political positions rather than resolve them is useful.

Comment author: gokhalea 01 November 2012 06:36:56PM 2 points [-]

Thanks, I tried to explain above. Less Wrong's conclusion on analyzing politics is flawed because it is based on the assumption that rationality with respect to politics requires an ideal answer. Pointing out that biases/emotions/etc. are ever present is used to protect the idea that rationality in its purest form always results in a normative answer. "Our model of rationality is always correct -- its just the people are flawed!!!" -- I disagree. The model is wrong. The people are playing their role as members of a social dynamic -- rationality in politics is dependent on their biases, not to be avoided because of them.

The value is awareness -- that is the true goal. To have an understanding of what is going on around you without confusion, anger, unwanted emotions. Rationality is about seeing the world "as it is." The world is social, and I want an understanding of how the world works, with its participants and their various viewpoints, perspectives, beliefs, and actions. I'm not trying to be "right" -- frankly i have political positions but don't really care -- they are a secondary concern to understanding the social dynamic.

Comment author: gokhalea 01 November 2012 06:22:58PM 2 points [-]

Thanks, and I appreciated Paul's article -- very interesting and insightful.

Let me try to clarify --

One of the issues causing confusion is that the definition of rationality is not commonly accepted/subject to some dispute. My understanding of EY's perspective on the definition of rationality is based on his article: What do we mean by rationality

EY is saying that applying rationality yields a normative answer -- and that LW is not receptive to a different idea, such as a model where an argument can be rational but still not be the "correct"/"true" answer. My argument is that rationality, as EY defines it, does not work with respect to politics because political issues do not have correct answers (i'll get to why shortly). So I don't disagree with your point that politics can be discussed rationally -- i just have a different definition of rationality when it comes to politics.

I read Paul's article -- it was very good -- i have previously considered the idea that in politics or religion, everyone is an "expert" and the idea of identities intertwined with people's positions -- no doubt insightful, but i think its incomplete. (i also note that his argument that politics has definite answers sometimes is baffling -- the cost of government policy is NEVER certain -- simply because people can't predict the future or how people will behave in the future).

The issue and uniqueness of politics is NOT that everyone is an expert -- its that everyone is a participant, in a real and legitimate way -- as a voter or policy maker or government leader. As such, politics is truly a social issue -- analytical analysis is possible, but you NEVER going to get a clear answer -- the social issues are forever intertwined with policy. Remember, regardless of how much weight you may put on ideal policies/laws/regulations, the ability of any leader to implement these policies is WHOLLY CONTINGENT on winning an election, thus drawing in all potential voters in the discussion/decision. Another way to think about this is trying to answer the question -- "how to be a good mother" -- this is a social issue among a mother and her kids within the context of their familial unit/environment. You may have high level guidance, but no one can answer this question -- its a dynamic issue that is forever unique in ways that can never yield an answer. I believe politics is the same.

Again, i think politics can be discussed rationally, but in a different context -- it should be analyzed like any other social issue. For example, when there is a personal conflict, there are theories on how to handle this -- you have an approach, but part of it depends on how the other person reacts, their positions, their biases, and WHY they have the particular perspective. RIght/Wrong is sometimes irrelevant because in social issues, being correct is a secondary concern to managing the social relationship (including biases/emotions/identities). Rationality is more subjective when it comes to politics -- and it is very possible to have two positions that are "subjectively" rational but contradict each other with respect to a particular issue -- in the same way you and your friend can disagree on whether you should study x or y or whether you should date a or b -- both can have valid arguments but ultimately a decision must be made. Focusing on the "right" answer is fruitless -- rationality is based on having the emotional intelligence to understanding the dynamics and uncertainty of this particular social relationship.

You may disagree, and thats fine -- I'm trying to learn and this is an exercise that is not easy -- however i point out that it provides an explanation for why rationality (as EY defines) has not yielded a clear answer and thus is a "mind killer." I think the model definition of rationality used here is simply wrong when applied to politics.

Comment author: Nornagest 31 October 2012 12:10:45AM 2 points [-]

Is that a forward flowing, rational conclusion? No one here can or will apply rationality in coming to political conclusions (whether a firm answer or not) -- so how can you tell me that its a mind-killer?

People here try to apply rationality to politics all the time. "Politics is the mind-killer" is an observation about its success rate.